271 research outputs found
Neural network based method for solving SMT problems
No full textSubmission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2025-10-19 without embargo termsThe student, Keyu Lu, accepted the attached license on 2025-05-01 at 20:25.The student, Keyu Lu, submitted this Thesis for approval on 2025-05-01 at 20:39.This Thesis was approved for publication on 2025-05-05 at 12:28.DSpace SAF Submission Ingestion Package generated from Vireo submission #22166 on 2025-10-19 at 18:11:31Satisfiability Modulo Theories (SMT) over nonlinear real arithmetic (NRA) represents a fundamental yet notoriously difficult problem class in formal verification and symbolic reasoning. Traditional SMT solvers struggle with the scalability and decidability of QF_NRA problems due to their intrinsic nonlinearity. In this work, we propose a novel reduction-based framework that translates SMT problems defined in the SMT-LIB2 format into equivalent neural network verification problems specified in the VNN-LIB format. By constructing tailored neural networks that capture the semantics of the original constraints, we leverage powerful neural network verifiers—specifically, the α,β-CROWN solver—to determine the satisfiability of the original NRA formulas. This transformation enables the application of recent advances in neural network verification to a broader class of symbolic problems. Our approach bridges the gap between symbolic logic reasoning and neural verification, potentially unlocking new paths for scalable and parallelizable SMT solving. We demonstrate the soundness and feasibility of the method through illustrative case studies and analyze its performance in terms of accuracy and approximation fidelity
When Fewer Layers Break More Chains: Layer Pruning Harms Test-Time Scaling in LLMs
No full textLayer pruning has emerged as a widely adopted technique for improving the efficiency of large language models (LLMs). Although existing methods demonstrate strong performance retention on general knowledge tasks, their effect on long-chain reasoning, a more brittle yet crucial capability, remains largely unexplored. In this work, we study the impact of layer pruning on long-chain reasoning through the lens of test-time scaling, a key mechanism in modern LLMs that enables strong reasoning capacity by allocating more computation at inference time. With extensive experiments, we demonstrate that pruning even one or two layers can severely impair test-time scaling, with performance collapsing drastically on long reasoning benchmarks even when performance on knowledge-intensive and shallow reasoning tasks remains stable. Furthermore, we find that standard supervised fine-tuning remedies fail to recover test-time scaling once it has deteriorated. Through in-depth analyses, we identify the mechanisms underlying this fragility of test-time scaling and highlight the fundamental risks of applying layer pruning to reasoning-intensive LLMs. These findings call for a rethinking of layer pruning strategies and provide insights for developing methods that preserve the robustness of reasoning. We open-source the codebase in \href{https://github.com/keyu-wang-2002/Layer-Pruning-Harms-Inference-Scaling}{https://github.com/keyu-wang-2002/Layer-Pruning-Harms-Inference-Scaling}
In Vitro Study of the Cytotoxic Effects of Low- and High-molecular-weight Fucoidan Extracted from New Zealand Seaweed Undaria Pinnatifida in MCF-7 and MDA-MB-231 Breast Cancer Cell Lines
No full textBreast cancer is known as the top cancer for women worldwide. It is estimated that every year over one million new cases of breast cancer are diagnosed and contribute largely to cancer related deaths. Chemotherapy, including neoadjuvant therapy and adjuvant therapy, is a critical part in treatment for breast cancer that impact on survival and life quality for patients. However, chemo-resistance and adverse effects occur frequently when patients receive chemotherapy or the improved target therapies. New strategies have been proposed to enhance the effects of anticancer drugs as combing them with natural dietary compounds, decreasing drug dose administered and reducing the toxicity to normal cells.
Fucoidan is noticed for its anti-cancer potential in treating breast cancer as well as in many other cancers. It is a natural bioactive compound derived from brown algae that has low toxicity and multiple anti-cancer pathways, the potential of which makes it a candidate for therapeutic agent using alone or in combination with other cytotoxic drugs. Base on the molecular weight, fucoidan can be categorised into three ranges: high-molecular weight fucoidan (HMF, >300k), medium-molecular weight fucoidan (MMF, 300-10k) and low-molecular weight fucoidan (LMF, <10k). In this study, the inhibitory effects of HMF and LMF from New Zealand Undaria Pinnatifida have been studied against breast cancer. Two breast cancer cell lines, MCF-7 and MDA-MB-231, have been used in this study representing ER-positive type and triple-negative type of breast cancer. A fibroblast (HDFa) cell line has also been used in this study, representing non-cancer cells, to examine toxicity of fucoidan.
By conducting MTT assays, apoptosis assay and other related mechanism assays on cancer cells, the findings in this study indicate that LMF exhibited much better inhibition on proliferation of breast cancer cells than HMF. Dose-dependent inhibition by LMF was observed in both MCF-7 and MDA-MB-231 after incubated for 48, 72 and 96 hours. MCF-7 cells are more sensitive to LMF than MDA-MB-231 by a distinction of about 20% inhibition at the highest concentration of LMF (56.6% inhibition at 200 µg/ml and 39.2% inhibition at 300µg/ml,72hrs, respectively) and time-dependent manner of inhibition was only observed in MCF-7. The IC50 of LMF to MCF-7 cells over 72 hours was determined to be about 19 µg/ml and dropped to 10.5 µg/ml after 96 hours. Induction of caspase-dependent apoptosis was observed in MDA-MB-231 cells through intrinsic apoptosis pathway alone or with the extrinsic pathway. An activation of NOS stimulated by LMF was observed in MDA-MB-231 cells at a dose-dependent manner. No obvious cytotoxicity of LMF to HDFa cells was observed by 72 hours incubation in a cell cycle assay. To conclude, LMF from New Zealand Undaria Pinnatifida showed great anti-cancer effects against these two types of breast cancer, therefore, it has great potential to be used as a therapeutic agent or a supplement to combine with other chemo-agents for treating breast cancer, even though it may not be potent enough to treat this type of cancer alone
In Vitro Study of the Cytotoxic Effects of Low- and High-Molecular-Weight Fucoidan Extracted from New Zealand Seaweed Undaria pinnatifida in MCF-7 and MDA-MB-231 Breast Cancer Cell Lines
No full textBreast cancer is known as the top cancer for women worldwide. It is estimated that every year over one million new cases of breast cancer are diagnosed and contribute largely to cancer related deaths. Chemotherapy, including neoadjuvant therapy and adjuvant therapy, is a critical part in treatment for breast cancer that impact on survival and life quality for patients. However, chemo-resistance and adverse effects occur frequently when patients receive chemotherapy or the improved target therapies. New strategies have been proposed to enhance the effects of anticancer drugs as combing them with natural dietary compounds, decreasing drug dose administered and reducing the toxicity to normal cells.
Fucoidan is noticed for its anti-cancer potential in treating breast cancer as well as in many other cancers. It is a natural bioactive compound derived from brown algae that has low toxicity and multiple anti-cancer pathways, the potential of which makes it a candidate for therapeutic agent using alone or in combination with other cytotoxic drugs. Base on the molecular weight, fucoidan can be categorised into three ranges: high-molecular weight fucoidan (HMF, >300k), medium-molecular weight fucoidan (MMF, 300-10k) and low-molecular weight fucoidan (LMF, <10k). In this study, the inhibitory effects of HMF and LMF from New Zealand Undaria Pinnatifida have been studied against breast cancer. Two breast cancer cell lines, MCF-7 and MDA-MB-231, have been used in this study representing ER-positive type and triple-negative type of breast cancer. A fibroblast (HDFa) cell line has also been used in this study, representing non-cancer cells, to examine toxicity of fucoidan.
By conducting MTT assays, apoptosis assay and other related mechanism assays on cancer cells, the findings in this study indicate that LMF exhibited much better inhibition on proliferation of breast cancer cells than HMF. Dose-dependent inhibition by LMF was observed in both MCF-7 and MDA-MB-231 after incubated for 48, 72 and 96 hours. MCF-7 cells are more sensitive to LMF than MDA-MB-231 by a distinction of about 20% inhibition at the highest concentration of LMF (56.6% inhibition at 200 µg/ml and 39.2% inhibition at 300µg/ml,72hrs, respectively) and time-dependent manner of inhibition was only observed in MCF-7. The IC50 of LMF to MCF-7 cells over 72 hours was determined to be about 19 µg/ml and dropped to 10.5 µg/ml after 96 hours. Induction of caspase-dependent apoptosis was observed in MDA-MB-231 cells through intrinsic apoptosis pathway alone or with the extrinsic pathway. An activation of NOS stimulated by LMF was observed in MDA-MB-231 cells at a dose-dependent manner. No obvious cytotoxicity of LMF to HDFa cells was observed by 72 hours incubation in a cell cycle assay. To conclude, LMF from New Zealand Undaria Pinnatifida showed great anti-cancer effects against these two types of breast cancer, therefore, it has great potential to be used as a therapeutic agent or a supplement to combine with other chemo-agents for treating breast cancer, even though it may not be potent enough to treat this type of cancer alone
A Remanufacturing News-vendor with Pricing and Take-back Pricing
No full textThis paper analyzes the problem of a remanufacturing news-vendor with selling and take-back price decision. In our model, the remanufacturer decides selling price, take-back price, and order quantity for new materials. She then uses the stochastic take-back quantity and the new material to meet the stochastic demand comparably to a news vendor setting. We allow demand and take-back supply to be correlated. In this thesis, we study a production problem with dual input sources: raw materials and recycled or remanufactured take-back items. To answer when mixed-sourcing is best, we analyze the model under deterministic setting first, provide criteria for different sourcing strategies, and give corresponding joint optimal solutions. Assuming that a mixed strategy is optimal, we then analyze the stochastic case, and find the optimal joint decision for raw-material order quantity, selling product price and take-back price. We find that, when the selling price remains fixed, the optimal take-back price and thus the expected take-back quantity does not change with increased demand and take-back supply variance. Also, the takeback price can exceed the net savings achieved by remanufacturing if consumers take this price into account when purchasing new products. And, the adding of randomness of demand and take-back supply will lower the optimal selling price and thus lower the take-back price. In future research, we will provide numerical analysis to report the impact and performance if a required recycling level is imposed in the problem; study the remanufacture problems with multiplicative demand function; multiple customer classes, such as the trade-in consideration; or multiple order opportunities, such as postponing the raw material procurement
A new model for understanding one-photon luminescence from single gold nanorods
No full textWe experimentally and theoretically studied the the photonluminescence from a single gold nanorod. Our theory explains the main features of the photon-luminescence radiation and is in good agreement with experimental observations. ? 2014 OSA.EI
- …
